new materials science andtechnology · 2015. 9. 1. · reprint from materials science andtechnology...

Reprint fromMaterials Science and Technology

A Comprehensive Treatment

Edited by R. W Cahn, P. Haasen and E. 1. Kramer

Chapt.er 7

Zirconium Alloys in Nuclear Applicationsby C. Lemaignan and A. T. Motta

from Volume 10 B

Nuclear Materials, Part 2edited by B.R.T. Frost

Published byVCR Verlagsgesellschaft mbHP. O. Box 101161, D-69451 WeinheimFederal Republic of Germany vc~

7 Zirconium Alloys in Nuclear Applications

Clement Lemaignan

CEA/Centre d'Etudes Nucleaires de Grenoble/DTP/SECC, Grenoble, France

Arthur T. Motta

Nuclear Engineering Department, The Pennsylvania State University,University Park, PA, U.S.A.

List of Symbols and Abbreviations 27.1 History. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 47.1.1 High Temperature Water Reactors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 47.1.2 Current Use. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 47.2 Fabrication and Products 57.2.1 Processing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 57.2.2 Microstructure...... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 77.2.2.1 Alloys and Alloying Elements " 107.2.2.2 Heat Treatments and Resultant Microstructure 177.2.3 Properties 187.2.3.1 Mechanical Properties 187.2.3.2 Diffusion Data 227.3 In-Reactor Behavior 247.3.1 Irradiation Damage and Irradiation Effects 247.3.1.1 Displacement Calculations 247.3.1.2 Irradiation Effects in the Zr Matrix 257.3.1.3 Irradiation Effects on Second Phases 287.3.1.4 Irradiation Growth 297.3.1.5 Irradiation Creep 327.3.1.6 Changes in Mechanical Behavior 337.3.1.7 Charged-Particle Irradiation 347.3.2 Corrosion Behavior 367.3.2.1 General Corrosion Behavior 367.3.2.2 Oxidation of the Precipitates 377.3.2.3 Water Radiolysis 397.3.2.4 Hydrogen Pickup 407.3.3 Pellet-Cladding Interaction 417.4 Challenges. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 467.5 Acknowledgements 477.6 References 47

2 7 Zirconium Alloys in Nuclear Applications

List of Symbols and Abbreviations

a,cao, CoA

IGSCCI-SCCLHGRLWRMIBKPCIPKAppmPWRRBMKRXR&DSCCSEMSIPASTEMSOCAP

RTBSTEMTrexUTSVVERYS

List of Symbols and Abbreviations 3

iodine intergranular stress corrosion crackingiodine stress corrosion crackinglinear heat generation ratelight water reactormethyl-isobutyl-ketone (process)pellet-cladding interactionprimary knock-on atompart per millionpressurized water reactorRussian graphite-moderated boiling water reactorfully recrystallizedresearch and developmentstress corrosion crackingscanning electron microscopestress induced preferential absorptionscanning transmission electron microscopesecond order cumulative annealing parameterstress relievedto be specifiedtransmission electron microscopytube-reduced extrusionultimate tensile strengthVoda-Voda energy reactor, Russian type PWRyield strength


7.1 History

7.1.1 High Temperature Water Reactors

Soon after the observation of the fissionof uranium 235, L. Szilard and F. Joliot-Curie recognized the possibility of usingthe chain reaction phenomenon as a sourceof energy. Initially, test reactors were de-signed with no constraints on thermal effi-ciency. The aim was then to understandneutron physics and to study the behaviorof materials under irradiation. Low tem-perature, pool type reactors were con-structed in which the structural materialused was exposed to a comparatively mildenvironment. Aluminum and beryllium al-loys were used for core components, due totheir low thermal neutron capture crosssection and acceptable corrosion rate inwater below 100°C.

Once nuclear power reactors for sub-marine propulsion and production of elec-tricity were designed, thermal efficiencybecame mandatory, and materials had tobe found that could withstand the hightemperature of the coolant, usually water.Zirconium (Zr), with its very low thermalneutron capture cross section, was a poten-tial candidate, but had poor ductility andcorrosion resistance. The first pressurizedwater reactors were loaded with fuel clad-dings and other structural elements (guidetubes and grids) made of stainless steel.

An improvement in neutron efficiencywas a driving force for the development ofindustrial type Zr-based alloys. At the endof World War II, the nuclear submarineprogram undertook a large effort in thatfield. Systematic testing and research anddevelopment (R & D) by the U.S. Navy re-sulted in the development of an efficienthafnium (Hf) separation process and in-dustrial scale ingot production procedures.During the test of a series of binary and

ternary alloys, an accidental contamina-tion of a Zr-2.5 % Sn (Zircaloy-1) melt bystainless steel caused the serendipitous dis-covery of an alloy of good corrosion be-havior. Composition variations aroundthis alloy led to Zircaloy-2. Zircaloy-3, avery low tin variant, was soon abandonedin favor of the better Zircaloy-4, aNi-freevariant designed to decrease hydrogenpickup. Similar tinkering with composi-tional variations to improve corrosion re-sistance and strength led to the develop-ment by the Soviet Union of anotherfamily of alloys using the Zr-Nb binarysystem, later used by Canada as well. TheZr-Nb system allowed the possibility ofobtaining a fine two-phase structure thatleads to higher strength.

The subject of Zr metallurgy has meritedbooks (Lustman and Kerze, 1955) and re-views (Douglass, 1971; Cheadle, 1975) inthe past. Detailed aspects of the metallurgyof the IV-A series (Ti, Hf, and Zr) can befound in Vol. 8, Chap. 8, of this Series. It isthe purpose of this work to review the useof Zr for nuclear applications and to pre-sent some of the more recent developmentsin the field.

7.1.2 Current Use

In today's nuclear power reactors, Zralloys are commonly used for structuralcomponents and fuel cladding. For lightwater reactors (LWR), the commonchoices are Zircaloy-4 in pressurized waterreactors (PWR) and Zircaloy-2 in boilingwater reactors (BWR). The heavy water-moderated natural uranium CANDU re-actor (Canadian deuterium uranium), aswell as the Russian RBMK reactor, useZr-Nb alloys.

In fuel assemblies and bundles, claddings are made out of Zircaloy-2 orZircaloy-4. Those components are exposedto the fission products at the inner surface

at temperatures close to 400°C. At theouter surface they are in contact with lightor heavy water at coolant temperatures(from 280 to 350°C). Typical heat fluxesacross the cladding are in the range of30-50 W· cm- 2. Those tubes have differ-ent geometries, depending on reactor de-sign (Fig. 7-1). In PWR's fuel rods clad-dings are 4 to 5 meters long and have adiameter of 9 to 12 mm for a thickness of0.6 to 2.8 mm. BWR fuel rods are usuallyslightly larger. In CANDU, fuel bundlesare short - 0.5 m - to allow on-line refu-elling. The cladding is very thin - 0.4 mm- and is designed to collapse around theV0 2 pellets early during irradiation. In theRussian VVER's the fuel rod geometry is3irnilar to PWR's but the usual claddingalloy is Zr-1 % Nb.

Structural components of the fuel as-semblies are guide tubes and grids thatcompose the skeleton. They have to with-stand mechanical stresses during normalor accidental operation as well as the oxi-dizing hot water. In BWR's each assemblyis surrounded by a Zircaloy-2 channel boxthat avoids cross-flow instabilities of thetwo-phase coolant. Geometrical stabilityof those components is a critical aspect ofcore design as it affects fuel loading capa-bility, cooling efficiency and neutronphysics behavior of the core.

In the case of CANDV's and RBMK'sthe coolant is separated from the modera~tor and flows around the fuel bundlesin pressure tubes, usually made of Zr-Nballoys. Those large components (10 m x20 cm x 5 mm) are considered as a struc-tural part of the reactor with a design lifeof tens of years. They are thus exposedto a high irradiation fluence (up to3 X 1026 n m - 2) in contact with thecoolant on the inner surface. Mechanicalstability of those large components affectsthe overall geometry of the reactor.

7.2 Fabrication and Products 5

7.2 Fabrication and Products

7.2.1 Processing

Zirconium is commonly found in natureassociated with its lower row counterpartin Mendeleev's table, hafnium. Most of thecommon Zr ores contain between 1.5 and2.5 % Hf. Due to its high thermal neutroncapture cross section, Hf needs to be re-moved from Zr for nuclear applications.

The most frequently used ore is zircon(ZrSi04 ) with a worldwide production ofabout one million metric tons per year.Most of the zircon is used in its originalform or in the form of zirconia (Zr02) asfoundry die sands, abrasive materials orhigh temperature ceramics. Only 5 % isprocessed into Zr metal and alloys.

The processing of Zr alloy industrialcomponents is rather complex due to thereactivity of the metal with oxygen. Thegeneral scheme is presented in Fig. 7-2:Ore processing, zirconium/hafnium sepa-ration, reduction to metal, alloy melting,hot and cold deformation processing.

The first step is to convert the zirconinto ZrCI4 , though a carbo-chlorinationprocess performed in a fluidized bed fur-nace at 1200 °C. The reaction scheme is thefollowing:

Zr02(+Si02+Hf02)+2C+2CI2~

~ ZrCl4 ( +SiCl4 +HfCI4) + 2 CO

After this step, Zr and Hf are separatedusing one of the two following processes:

(i) Wet chemical: after reaction withammonium thiocyanate (SCN NH4 ) asolution of hafnyl-zirconyl-thiocyanate(Zr/Hf)O(SCN)2 is obtained. A liquid-liquid extraction is performed with meth-yl-isobutyl-ketone (MIBK, name of theprocess). Hf-free Zr02 is obtained afterseveral other chemical steps: hydrochlona-tion, sulfation, neutralization with NH 3 ,


(b)

(a)

(c)

Figure 7-1. Fuel cladding and other components,made of Zr alloys, used in different reactor types: (a)PWR fuel assembly (courtesy FRAGEMA), (b) BWRfuel assembly and channel (courtesy GEc), (c)CANDU fuel assembly and surrounding pressuretube.

-

and calcination. ZrCl4 is the final resultof a second carbo-chlorination process(Stephen, 1984).

(ii) Direct separation process: this is anextractive distillation within a mixture ofKCI-AICI 3 as solvent at 350°C. The vaporphase, generated by a boiler at the lowerpart of the distillation column, is enrichedin Hf, while the liquid phase traps the Zr(Moulin et aI., 1984b).

In either case, Zr metal is obtained by areduction of ZrCl4 in gaseous form byliquid magnesium, at about 850°C in anoxygen-free environment. Residual quan-tities of Mg and MgCl4 are removedfrom the "sponge cake" by distillation at1000°C. After mechanical fracturing, the)ieces of sponge are sorted, giving the ba-sic product for alloy ingot preparation.

High purity Zr can be obtained by theVan Arkel process. This consists of the re-action of Zr with iodine at moderate tem-perature, gaseous phase transport as ZrI4and decomposition of the iodide at hightemperature on an electrically heated fila-ment, the iodine released being used for thelow temperature reaction in a closed looptransport process, according to the follow-ing scheme:

Zr+2Iz~ ZrI4 (g) ... ZrI 4 ~ Zr+2Iz(g)t I

250- 300°C 1300-1400°C

For industrial alloys, a compact ofsponge containing the alloying elements -° (in the form of Zr0 2), Sn, Fe, Cr, Ni,and Nb - in the desired composition, ismelted in a consumable electrode vacuumfurnace, usually three times. These vacuummeltings reduce the gas content and in-crease the homogeneity of the ingot. Typi-:al ingot diameters range between 50 and

80 cm, for a mass of 3 to 8 metric tons.Industrial use of Zr alloys requires either

tube- or plate-shaped material. The first


step of mechanical processing is forging orhot rolling in the ~ phase, at a temperatureclose to 1050 °C. Hot extrusion is used toobtain tube shells or Trex (tube-reducedextrusion), while hot rolling is used for flatproducts. For Zircaloys, at that stage a ~quench is performed to increase the corro-sion resistance of the final product. Thistreatment controls the distribution of sec-ond phase particles, if no further process-ing is performed above 800°C (Schemel,1977). Further reduction in size is obtainedby cold rolling either on standard orpilger-rolling mills. Low temperature re-crystallization is performed between thevarious size reduction steps.

7.2.2 ~crostructure

Pure zirconium crystallizes at ambienttemperature as an hexagonal close-packedmetal, with a cia ratio of 1.593 (i.e., a slightcompression in the c-direction comparedto the ideal ratio of 1.633). Lattice parame-ters are ao=0.323 nm and co=O.515 nm(Douglass, 1971). The thermal expansion'coefficients have been measured by Lloyd(1963) on single crystals. The difference inthermal expansion coefficients between thea and c-directions (see Table 7-1) impliesthat the cia ratio tends towards the idealratio at higher temperatures - i.e., towardsa more isotropic behavior.

At 865°C, Zr undergoes an allotropictransformation from the low temperatureh.c.p. (J, phase to body centered cubic~ phase. On cooling, the transformation iseither martensitic or bainitic, dependingon the cooling rate, with a strong epitaxyof the (J, platelets on the old ~ grains ac-cording to the scheme proposed by Burg-ers (1934) :

(000l)a;11{110}13 and


Chloridedistillationprocess

Chlorination

Carbon

Zircon(Si02 • Zr02 .Hf02 )

Zr/HfSeparation

Hafnium-free ZrCI,T

Compacted pure ZrCI,

T

TY

Inspection ~

T

Figure 7-2. Processing andfabrication showing all thesteps in the fabrication ofZr alloy components (cour-tesy CEZUS): (a) From zir-con to sponge Zr, (b) fromsponge to alloy ingot, (c)from ingot to final product.

Vacuum distillationZr.Mg _ Zr + Mg

Zirconiumsponge

Kroll reductionMgCI2 ZrCI, + Mg - Zr. Mg + MgCI2-+

Mg...

Magnesium-+

Crushing

Blending

(a)

metal. The main properties of the Zr and Zralloys are given in Table 7-1. It should benoticed that the main reason for selectingZr as a nuclear material is its low thermalneutron capture cross section, which isabout 30 times less than that of iron, givinga better thermal reactor neutron efficiency.

One should also note its strong aniso-tropic behavior. For elastic properties,the differences in thermal expansion andYoung's modulus along the main directionof the hexagonal lattice induce the develop-ment of internal stresses after any heattreatment due to grain-to-grain thermal

-


First SecondRaw melt (third)materials melt

Electrode Ingot

::::':b,.E~:.~i:9~ ·~r4,EI.::dj:~~~~l/'"'1scrap ~ ~ ~ + WaterAlloy ~ Electrode Arc outadditives I!!!!!I Compacting welding, Liquid

Water-cooled t poolcopper Water

(b) crucible in

Forgingpress Hollow billet

Wire in coil

--......~ Tubing

Coil

~o

Billet

Cold rolling

Sendzimir cold-rolling mill

Wire rolling mill

)dpen- or close-

dIe forging

! 1-. Forged partsBar

C~ •Machined partsWire drawing


strain incompatibilities; after annealing at500°C, the (c)-planes are thus in tensionat stresses up to 85 MPa, depending on theoriginal texture (MacEwen et al., 1983). Ina similar way, after plastic strain, elas-tic recovery is orientation-dependent andleads to compression along the (c) planesof almost 200 MPa (Holt and Causey,1987). For an industrial material, the elas-tic and thermal expansion coefficientshave to be computed from its texture, de-scribed either by the pole figures (Rosen-baum and Lewis, 1977) or better by usingthe complete orientation distribution func-tions (Sayers, 1987).

The relative solubility of the various al-loying elements in the II and ~ phases is oneof the bases for the choice of additions aswell as heat treatments.

7.2.2.1 Alloys and Alloying Elements

The zirconium alloys in use today fornuclear applications are limited in num-ber: besides pure Zr, only four alloys arecurrently listed in the ASTM standards(ASTM, 1990). Those are shown in Table7-2. The first three are used for claddingand structural materials, like guide tubesin PWRs and BWRs, channel boxes inBWRs and structural materials in CANDUreactors, while the last one, grade R 60904,is used exclusively in pressure tubes forCANDU reactors.

For cladding tubes, only Zircaloy-2 and4 are listed in ASTM B 811-90. Other alloyshave been developed during the history ofnuclear power, but except for the Zr-1 0/0Nb alloy used for cladding in RussianPWR's (VVER) (Tricot, 1990), none are incurrent use anymore, except for specializedapplications such as the Zr-Nb-Cu alloyused in garter springs for CANDU pres-sure tubes.

The needs for better performance of nu-clear fuel assemblies and structural parts,mainly with regard to corrosion resistance,has led metallurgists and fuel designers tointensive R&D efforts in order to improvethe properties of those Zr alloys by ad-vanced compositions and thermomechani-cal processing, and to optimize the mi-crostructure within the current ASTMalloy specifications.

Indeed, although safety concerns restrictthe introduction of new alloys because ofthe large amount of data that needs to beaccumulated to verify the safe behavior offuel elements in case of reactor accidents,the specifications for the alloys used todayare broad enough for optimization ofproperties within the specified compos.tion ranges. Moreover, the microstruc-tures may be varied significantly becauseof the ll-B phase transformation of zirco-nium, and because of the different solubil-ities of the alloying elements in the differ-ent phases. The main alloying elements arenow considered in turn:

Oxygen is to be considered as an alloy-ing element, and not an impurity. It isadded to the compacts before melting assmall additions of Zr0 2 powder. The usualoxygen content is in the range of800-1600ppm and its purpose is to increase the yieldstrength by solution strengthening. A 1000ppm oxygen addition increases the yieldstrength by 150 MPa at room temperature,(Armand et al., 1965). Oxygen is an II sta-bilizer, expanding the II region of the phasediagram by formation of an interstitialsolid solution.

The Zr-O phase diagram is given inFig. 7-3: at high concentration, oxygenstabilizes the II phase to liquid temperatures: During high temperature oxidation,simulating a reactor accident, a layer ofoxygen-stabilized ll-zirconium is found be-

-


I Table 7-2. Composition range of standard Zr alloys.ASTM Ref. R 60802 R 60804 R 60901 R 60904

Common name ZircaIoy-2 Zircaloy-4 Zr-Nb Zr-Nb

Alloying elements (mass %)

Sn 1.2 -1.7 1.2 -1.7Fe 0.07-0.2 0.18-0.24Cr 0.05-0.15 0.07-0.13Ni 0.03-0.08Nb 2.4 -2.8 2.5-2.80 to be specified on order usually 1000-1400 ppm 0.09-0.13 TBS

Impurities (max. ppm)

Al 75 75 75 75B 0.5 0.5 0.5 0.5Cd 0.5 0.5 0.5 0.5C 270 270 270 150Cr 200 100Co 20 20 20 20Su 50 50 50 50Hf 100 100 100 50H 25 25 25 25Fe 1500 650Mg 20 20 20 20Mn 50 50 50 50Mo 50 50 50 50Ni 70 70 35Ni 80 80 80 65Pb 50Si 120 120 120 120Sn 50 100Ta 100Ti 50 50 50 50U 3.5 3.5 3.5 3.5V 50W 100 100 100 100

tween the ~ quenched structure and thezirconia.

Tin is also an a, stabilizer. It forms in thea, and ~ phases a substitutional solid solu-tion. Tin-based precipitates have been re-ported in the literature (Bangaru, 1985)but they appear to be artifacts of TEMsample preparation (Charquet and Alheri-tiere, 1985). At a concentration of 1.2-1.8%, it is used for an increase in corrosionresistance especially by mitigating the dele-terious effect of nitrogen in deteriorating

corrosion behavior. Due to a better con-trol of processing parameters, and conse-quently of nitrogen content, the usage oftin tends to be lower in the current alloys.Tin, however, also has a limited impact onmechanical properties, by increasing thetensile yield strength, and therefore itscomposition should not be excessively re-duced.

Iron, chromium, and nickel are consid-ered as "~-eutectoids", because, in theirphase diagrams, these elements have an

-


oWeight Percent Oxygen

5 10 15 20 25 30 35

Figure 7-3. Zr-O phasediagram (Abriata et aI.,1986).

70

.. ,I I

271 O°C :..~.t.G....... ,L

Q~f:-::2.~~~:~~:

2065°C---F=::::""4LO-...=.::..:.=.-=-----:6~2 :

• I• I

__ _:_1_~~~~~. ..~~~~ :__ :31.2 66.51

-12050 C BZr02_X._---. __ ._.----------- .. -.---------- -------

29.8-- . :: ~?P.o~ .. .. _---_. _...

(aZr)

25%,2130°C

2800

2600

2400

2200

0 2000

~ 1800

~ 1600~ 1400~ 1200

1000:' 29.1 66.7:

800 86.3°C (a' Zr) :' : aZr02_x

600 ( "Z) (C!:l" zr).~ ••••• __ • :-_qQQ~9 .1400 ~ r .~': ..... 28.6 66.7:

(al"Zr),~....¥··\\ \7 (a4"Zr) :200 -+------T-'-'''''-----'-r-''-'--'-'--'-.------,-------,--__;---'--...,-----'o 10 20 30 40 50 60

Atomic Percent Oxygen

eutectoid decomposition of the ~ phase(Fig. 7-4 to 7-6). As mentioned in Sec.7.1.1, they were added to the early Sn-based alloys after an accidental pollutionby stainless steel of a melting lot showed anenhancement in corrosion resistance, lead-ing to the Zircaloys 2 and 4.

At common concentrations, these ele-ments are fully soluble in the ~ phase. Thetemperature of dissolution of those ele-ments is in the range of 835-845 °C - thatis, in the upper r:t + ~ range (Miquet et al.,1982). In the r:t phase their solubility is verylow: in the range of 120 ppm for Fe and

1009080

Weight Percent Iron

o 10 20 30 40 50 60 702000+---'--,--'----+---',--------'--,-----'--.----'-r------,L--.---L--,--------i

Figure 7-4. Zr- Fephase diagram (Ariasand Abriata, 1988).

100

99.92

(aFe)-

90

1538°C(oFe) ,

-92.9,.'.. 1394°C

8070

90.2 -99.3:, (yFe)~

1357°C

; 92_5_oC ---i912°C

: no°c~gneticTransformation:-ZrFe3..y 470°C

" 1482°C-r-------, ," 1337°C

60

~'I>~~~\'Ii: -2750C

,--------: Magnetic Transformation

50

55.1

403020

L

10

200

1855°C1800 ,

", '1600 ~ \, ,, ,, ,1400 , ,

o "';"(BZr)' ,

i :::: 6\928j~.jl> m ~':4:?------- --~ 863°C r: - - - -;.r.- -,7- Zr2Fe cD~ 800 '- I' 87835;~ ;',. _-'~ ?C~O~_ _____ LL~ 4.0 :: N

600 0.02 ::

-(a"Zr) :;400 : : _3000C

~::---------

N ::o+-_---,-----J'L..:-.-----,------,--,---'--..,.---'---


Weight Percent Chromium

o 10 20 30 40 50 60 70 80 90 1001900 +-18-55-

o-C.----L--.-----+--,---L------,--L----,r-'----.-'--+--L-'1-86-"-3-oC

-I

Figure 7-5. Zr-Crphase diagram (Ariasand Abriata, 1986).

90 100Cr

80

yZrCr2 66.7%,1673°C 0

..... ----~.:: _\19?. ---. 15920 C 1

------.1-~~71t-1- _ -:~.~ 15320Ct-~~:~~BZrCr2 : ~

:~ (Cr)-.l[)

1332°C : ,....ir---(0

30 40 50 60 70Atomic Percent Chromium

L

20700 +-----.-----,---,-------r--.-----.------'-------',--..,--------,----j

o 10Zr

900

1100

1700

2: 1500~:::l

"§(1)

~ 1300

~

200 ppm for Cr at maximum solubilitytemperature (Charquet et aI., 1989 a). Forthe Zr-Cr and Zr-Ni binary alloys, the sta-ble forms of the second phase are Zr2Ni orZrCr2' These phases are effectively theones observed in the Zircaloys, with Fesubstituting for the corresponding transi-

tion metal. The Zr3Fe phase which ap-pears in the binary Zr-Fe diagram is notfound in Zircaloy, probably because itsformation is too sluggish (Bhanumurtyet aI., 1991).

Therefore the general formulae of theintermetallic compounds in Zircaloy are

Weight Percent Zirconium

o 10 20 30 40 50 60 70 80 90 100

1800

1600 L

1455°C 1440°C0 1400e.....~:::l

~Q)

@- 1200~

1000 960°C

76(BZr)

845°C

800(aZr)

0 10 20 30 40 50 60 70 80 90 100Ni Atomic Percent Zirconium Zr

Figure 7-6. Zr- iphase diagram ( ashand Jayanth, 1984).

-


Zr2(Ni, Fe) and Zr(Cr, Fe)2' In Zircaloy-4,the Fe/Cr ratio of those precipitates is thesame as the nominal composition of thealloy. In Zircaloy-2 alloys, the partitioningof Fe between the two types of intermetal-lic phases leads to a more complex rela-tionship between nominal compositionand precipitate composition, giving abroad range of Fe/Cr ratio in Zr (Cr, Fe)2'and Fe/Ni in Zr2(Fe, Ni) (Charquet andAlheritiere, 1985; Yang et aI., 1986).

The crystal structure of the Zr(Cr, Fe)2precipitates is f.c.c. (C 15) or h.c.p. (C 14),depending on composition and heat treat-ment, with characteristic stacking faults asseen in Fig. 7-7. Both structures are Lavesphases. The equilibrium crystallographicstructure is dependent upon the Fe/Cr ra-tio, cubic below 0.1 and above 0.9, andhexagonal in the middle, following an em-pirical rule proposed by Shaltiel et aI.(1976). In common alloys, both types ofstructures are found, even in the samesample, with random probabilities of oc-currences of each. The Zr2(Ni,Fe) precipi-tates have a body-centered tetragonal C 16structure (AI2Cu-type).

The size of these precipitates is of impor-tance for the properties of the alloys, espe-cially the corrosion rate: while better uni-

form corrosion resistance is obtained forZircaloys used in PWRs if they containlarge precipitates, better resistance to lo-calized forms of corrosion is seen in BWRsin materials that have finely distributedsmall precipitates. It has been shown thatprecipitation after ~ quenching is rapid(less than 10 min at 500°C) and that thecoarsening rate of those precipitates con-trols their sizes in the final microstructure.It is thus necessary to consider the com-plete history of the various heat treatmentsfollowing the final ~ quenching to assessthe final precipitate size distribution. Var-ious cumulative annealing parameters(CAP or LA) have been proposed for thispurpose, based on recrystallization activa-tion energy or corrosion behavior. Recently, the coarsening kinetics have beenshown to be second order with an activa-tion energy of Q/R = 18 700 K, and a betterdescription of the resultant precipitate sizedistribution results from using the secondorder cumulative annealing parameter(SOCAP) (Gros and Wadier, 1989).

In the Zr-2.5 % Nb alloy used inCANDU pressure tubes, very little Fe isfound in the C1 phase, most of it being in theremanent ~ phase, in metastable solid solu-tion.

200 nm

Figure 7-7. Typical distri-bution of second phaseprecipitates Zr(Cr, Fe)2 andZr2 i, Fe) in recrystallizedZircaloy.

-


Figure 7-8. Zr-Nb phasediagram (Abriata andBolcich, 1982).

Weight Percent Niobium

10 20 30 40 50 60 70 80 90 100

10 20 30 40 50 60 70 80 90 100Atomic Percent Niobium Nb

I

2469°e-~

L /--------/1855°e~ ------~}7L ooe ,-' - -,---

,-'- -21.7

(BZr, BNb)

988°C

863°C -------- 60.6 ~r--..r-........~ V I ~620°C~(aZr) 1 .5 91.0

800

600

400o

Zr

~ 1600'E03 1400D-E~ 1200

1000

o2600

2400

2200

2000

01800~

Niobium (columbium) is a ~ stabilizer.From pure ~-Zr to pure Nb there exists acomplete substitutional solid solution athigh temperature (Fig. 7-8). A monotec-toid transformation occurs at about 620°Cand around 18.5 at. % Nb. By waterquenching from the ~ or upper r:J. + ~ re-gions, the ~ Nb-rich grains transform bymartensitic decomposition into an r:J.' su-persaturated h.c.p. phase; subsequent heattreatment below the monotectoid tempera-ture leads to the precipitation of WNbprecipitates at twin boundaries of r:J.'needles (Williams and Gilbert, 1966). Inaddition a metastable co phase can be ob-tained from the ~ by slow cooling or agingof a quenched structure. A simple epitaxialrelationship is obtained between the cophase and the parent ~ (Dawson and Sass,1970).

Hydrogen is not an alloying element byeesign, but its behavior has to be assessed,

since during waterside corrosion, both thehydrogen produced by the reduction of thewater for oxidation of the Zr matrix and

the hydrogen present for water chemistrycontrol, can be absorbed to some extentinto the bulk of the alloy. Hydrogen atomsare located at tetrahedral sites of the h.c.p.cell of the Zr matrix up to the solubilitylimit (about 15 ppm at 200°C and 200 ppmat 400 DC). Above the solubility limit hy-drogen precipitates as the equilibrium ()f.c.c. phase (ZrH1.66)' The correspondingphase diagram is given in Fig. 7-9. Highcooling rates cause the precipitation of themetastable body centered tetragonal yphase, ZrH (Weatherly, 1981).

Due to the volume expansion inducedby the precipitation of the hydrides, thisnew phase tends to reduce its strain energyby nucleating on low index crystallo-graphic planes. Habit planes are {10tO} forpure Zr and {1 Ot7} for Zircaloys, inepitaxy with the matrix according to therelationship (111)0 II (0001hr (Bradbrooket aI., 1972). Further macroscopic growthof the hydride clusters occurs in the planeof maximum tensile stresses or in the basalplane if unstressed (Kearns and Woods,


Hybride

8-E

I'" .'" ..'"oe.•

• + 0 •8 - Hybride 0

oo

Zr (B)

Zr(a)+

Zr(B) o........~-..........----".....'----.e---- -0- -', 0

Zr (E) + 8 - Hybride (Cubic) 0 \ 00

Zr(a)

950.---------------------------,

900850800

750~ 700~ 650::J

Ci5 600~ 550E~ 500

450400350300 l--- -----J

50 : 8: E-HybrideZr (E) + 8 - Hybride (Cubic) 8 : ~:(Tetragonal)

O~--L-----.---------.-____._-_____,--._-__.__-__.__----'--___+_'::""O"":"'___,__--=---_____r'_'

o 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0H/Zr

"'.ott.

Atomic Percent Hydrogen

0 10 20 30 40 501000

900

/863°C800 (BZr)0°i 700:J

~ 600 550°CQ) -'-n. 0.07 -0.659°C 1.43°CE 500~

(aZr)400

300

200

100

00 0.2 0.4 0.6 0.8 1 1.2 1.4Zr Weight Percent Hydrogen

60

I)

;""" E

": :, ,

Figure 7-9. Zr-Hphase diagram (Zuzeket aI., 1990).

1.6 1.8 2

1966). Thus the texture of the material aswell as the stress state are critical parame-ters for the control of the precipitationmorphology of the hydrides, a brittle con-stituent at low temperatures.

Other minor constituents are often foundin the form of precipitates. Among themare the carbide f.c.c. ZrC and silicides cphosphides of various stoichiometries(Zr3Si, ZrSi2 , ZrP, Zr3P) (Charquet andAlheritiere, 1985).

-

7.2.2.2 Heat Treatments and ResultantMicrostructure

At high temperature, the oxide layerthat develops during forging is not protec-tive and therefore 0, N, and H can diffuseinto the bulk of the ingot if no particularcare is taken.

Some of the precautions that have to betaken for thermomechanical processingare reducing the time at high temperaturein unprotected atmospheres and perform-ing intermediate descalings and etchings ofoxidized forgings.

Thus, after ingot melting, the ther-momechanical processing commonly usedfor industrial alloys is the following:• Hot forging in the ~ range (1000 to

1050°C).• Water quenching from the homoge-

neous ~ phase (above 1000°C).• Intermediate temperature (upper Ct)

forging and rolling, or extrusion fortubes.

• A series of cold temperature roilingsfollowed by intermediate anneals invacuum furnaces.Homogenization in the ~ phase leads to

the complete dissolution of all the secondphase particles, but gives rise to significantgrain growth: after 30 min at 1050°C,grain size may reach several millimeters.During the water quench, the ~ grainstransform into Ct needles by bainitic trans-formation due to the slow cooling rate ofthe large pieces involved. These Ct needlesnucleate at grain boundaries and each for-mer ~ grain leads to a series of crystallo-graphic orientations corresponding to the12 different permutations of the Burgersrelationships. This leads to a typical "bas-ket-weave" microstructure as seen inFig. 7-10 a. The ~-eutectoid elements arerepelled by the transformation front andprecipitate at the boundaries of those


(o)

(b)

Figure 7-10. Typical microstructure of pquench ma-terial: (a) Basket weave l1. grains from the same former~ grain. The contrast obtained in polarized lightshows the 4 different orientations of the l1. grains fromthe same ~ grain. (b) Second phase precipitation at l1.grain boundaries. In bright field, the precipitates ap-pear as small white dots at interplate boundaries.

needles (Fig. 7-10b). This ~ quench is usedfor an increase of corrosion resistance ofthe final product, and is a reference statefor further processing. The cold workingsteps increase the homogeneity of the pre-cipitate distribution.

After each cold working step of plate ortube material, an annealing treatment ismandatory to restore ductility. It is usuallyperformed at 550-600°C to obtain thefully recrystallized material (RX). The re-sultant microstructure is an equiaxed ge-ometry of the Zr grains with the precipi-


tates located at the grain boundaries andwithin the grains (Figs. 7-7 and 7-11).

To improve the mechanical properties ofthe final product, the temperature of thelast annealing treatment can be reduced toavoid complete recrystallization. This isthe stress-relieved (SR) state, characterizedby elongated grains and a high density ofdislocations.

In the case of Zr-2.5 % Nb alloys, ~quenching in water of small pieces leads tothe precipitation of at martensite supersat-urated in Nb. Tempering at intermediatetemperature results in ~-Nb precipita-tion at the lath boundaries and at twinboundaries within the lath (Williams andGilbert, 1966), followed by transformationof at into a. When quenching is performedfrom an a + ~ region, a uniform distribu-tion of a and ~ grains is obtained, and theNb-rich ~ phase does not transform. Inthis later case, however, the texture is lessuniform along the length of the tube thanin ~ quenching. After rolling or extrusion,the Nb-rich ~ grains tend to align and theresultant microstructure is shown in Fig.7-12. By aging, at temperatures in therange of 500°C, the metastable Nb-rich ~phase can be decomposed into an h.c.p. co

Figure 7-11. Microstructure of recrystallized Zircaloy-4: equiaxed a. grains with homogeneous distributionof intermetallic precipitates.

phase. This gives a sharp increase in me-chanical strength due to the fine micro-structure obtained by the ~-HD transfor-mation (Cheadle and Aldridge, 1973). Inthe usual form of the Zr-2.5 % Nb, the coldwork condition after a + ~ extrusion andair cooling, the microstructure consists ofZr grains with layers of Nb-rich ~ phase(close to eutectoid composition). Due tothe affinity of Fe for the ~ phase, most ofthis element is found in the minor ~ grains.These ~ grains are metastable and de-compose upon aging to a mixture of a-Zrand pure ~-Nb. The a-Zr phase itself ismetastable and irradiation induced precip-itation of the supersaturated Nb solid solu-tion can occur, which is believed to im-prove corrosion resistance (Drbanic et al.,1989).

In the case of Zr-l % Nb used for VVERand RBMK, the concentration of Nb islow enough to avoid microstructural evo-lution during service.

7.2.3 Properties

7.2.3.1 Mechanical Properties

The mechanical properties of the Zr al-loys are strongly dependent on severalparameters such as composition, textureand metallurgical state. For practical pur-poses, the properties at 300-400°C aremost important and room temperature be-havior is used mostly for comparison.

As reviewed by Tenckhoff (1988), thedeformation mechanism of the hexagonalZr follows two main mechanisms, slip ortwinning, depending on the relative orien-tation of the grain in the stress field.

Dislocation slip occurs mostly on prismplanes in the a-direction. This is referredto as the {10IO}

(1120) RADIALDIRECTION

(1010)LONGITUDINAL ----


Zircaloy-2Rolled Plate

(0002) POLE FIGURE

R

Figure 7-13. Pole figure of 1.5% cold-rolled Zirca-loy-2 sheet, showing the distribution of basal planesat 35 DC from the normal direction, and along theshort transverse direction in the sheet. R indicates therolling direction and T the short transverse. The nor-mal direction is perpendicular to the page. The thickcontour line indicates a random concentration ofbasal poles, the dotted line is half of random, and eachof the thin contour lines indicates an increase of halfthe random value, so that the maximum correspondsto about five times the random value. (Courtesy ofJohn H. Root, AECL, Chalk River Laboratory.)

After cold processing the (1010)-direc-tion is parallel to the rolling direction.During recrystallization heat treatment, a30 degree rotation occurs around the c-di-rection and the rolling direction is thenaligned with the (1 I20)-direction for someof the grains.

At room temperature, in the annealedstate, pure, oxygen-free Zr, has a low yieldstrength of 150 MPa. This yield strengthcan be enhanced by solution strengthen-ing, using alloying elements that have sig-nificant solubilities in cx-Zr. Oxygen, tinand niobium are candidate elements to

be considered. Nitrogen would be efficient,but degrades the corrosion resistance.Tin causes only a small increase in ten-sile strength (Isobe and Matsuo, 1991).By contrast, the addition of 800 ppm ofoxygen increases the yield strength to300 MPa. As a result of this, Zircaloyshave minimal yield strengths in the rangeof 250-300 MPa and the Zr-2.5 % Nb al-loy, 300 MPa. As in other metals, reduc-tion in grain size is also used to obtainhigher strength, leading to the specifica-tion of a grain index of 7 or finer for stan-dard products. For all those materials, theductility remains high (above 200/0). Addi-tional strength is obtained by cold work-ing, allowing the increase of the yieldstrength above 400-450 MPa. This is fo 'lowed by a final stress-relief heat treatmentto restore ductility without drastic reduc-tion in strength.

Finally the texture itself can increase al-loy strength by changing the Schmid factorfor slip or twinning. This can be observedby the differences in strength between theaxial and transverse directions. In addi-tion, due to the distorted shape of the yieldlocus, and consequently of the orientationof the strain vector, strain is also aniso-tropic.

The resistance to unstable crack growth(fracture toughness) is relevant to largecomponents under tensile stresses. There-fore, the fracture behavior of the Zr alloyshas been studied only in the practical caseof the large pressure vessels for chemicalengineering (Tricot, 1989) or pressuretubes in CANDU reactors. For hydrogen-free Zr alloys there does not exist any brit-tle-ductile transition with temperature asin ferritic steels and the rupture is alwaysductile. The embrittlement of Zr alloys'related to the behavior of hydrogen. Dur-ing operation, some hydrogen pick-up oc-curs (see Sec. 7.3.2.4). This hydrogen pre-

cipitates in the form of platelets havinga shape which is a function of textureand stress state. It can also diffuse alongthe thermal gradient, giving locally muchhigher concentration of hydrides. Thosehydrides are not ductile at low tempera-ture, and the fracture behavior of an alloycontaining significant amount of hydrogenis strongly dependent on temperature andhydrogen content. The geometrical distri-bution of the hydrides has also a large in-fluence: in the case of hydrides parallel tothe macroscopic plane of the crack, crackgrowth is enhanced by the percolation ofthe fractured platelets (Asada et aI., 1991).The values reported for the fracture tough-ness K1c of different alloys lie in the same_'ange of 120-150 MPa' m I/2 for hydro-gen free or for high temperature data. Thetransition temperature is reported around250- 300°C for 200 ppm of hydrogen anddecreases with hydrogen content. For hy-drogen content above 400 ppm, the roomtemperature fracture toughness is reducedbelow 30 MPa . m I/2 (Simpson and Chow,1987; Asada et aI., 1991). A reduction ofthe fracture toughness of Zr-2.50/0 Nbpressure tubes with the neutron irradiationfIuence has been observed and can belinked to the change in microstructure (dis-location density and Nb-rich precipitates).

Thermal creep: The stress applied tostructural parts under reactor conditionsis sufficient to induce creep. Although in-reactor creep is always mixed with growth(Sec. 7.3.1.4), some experiments were con-ducted to identify the basic mechanisms ofcreep strain out-of-flux. Compared withmetals of similar melting temperature, thethermal creep rates of zirconium alloys arehigh and design limitations result from this)articular behavior.

Zirconium creeps at low temperature: atroom temperature, grain boundary slidingis observed at low strain rates. Early creep


experiments on pure Zr have shown a largevariation of the activation energy withtemperature, indicating numerous mecha-nisms involved in this deformation pro-cess, that have not been clarified (Guibertet aI., 1969). In the range of reactor tem-peratures, the activation energies of 2.7 eVfor standard alloys are close to Zr self dif-fusion values (see Sec. 7.2.3.2). The stressexponents of the creep rate are still underdiscussion, reported to be in the range of 2at high stresses (Matsuo, 1987) or muchhigher at low stresses (Murty and Adams,1985). The mechanisms considered arecomplex with predominance of dislocationglide controlled by local climb.

Due to the texture of the cylindricalcomponents and to the different deforma-tion processes that could be considered,depending of loading path, analyticalcreep experiments are based on biaxialloading: closed end samples are tested intension or compression under variousinternal pressures. Axial and diametralcreep strains are measured in order to plotthe creep stress locus. Testing stress-re-lieved (SR) and fully recrystallized (RX)Zircaloys, Murty and Adams (1985) haveshown that the results are best explainedif prism slip is considered as the activemechanism for RX material and basal slipfor SR.

The response of Zr alloys to abruptchanges in loading stress level during creeptesting has been discussed in detail and aconsensus exists that a strain hardeningrule should be used for moderate changesin applied stresses (Lucas and Pelloux,1981). In the case of large changes of ap-plied stress for reversal loading conditions(tensile/compression) some contributionof recovery has to be considered (Matsuo,1989).

Most of the metallurgical parametersaffect the creep rate. For example, while


Zircaloy-2 and Zircaloy-4 have similarcreep strengths for the same thermome-chanical processing, Zr-2.5 % Nb is muchmore resistant to creep. The metallurgicalstate of the material also influences thecreep mechanisms: although a SR materialhas higher tensile strength, its high disloca-tion density allows it to creep about 2 to 3times faster than the RX material.

The effect of alloying elements on creepproperties has been found to be slightlydifferent from their effect on tensilestrength. The effect of oxygen remains im-portant to improve creep resistance, but itis smaller than the room temperature yieldstrength improvement. Although it haslittle effect on tensile strength, tin increasesthe creep resistance (McInteer et aI., 1989).Carbon also reduces the creep rate, butdue to its low solubility (150 ppm) it can-not be used efficiently for that purpose.Irradiation creep is treated in Sec. 7.3.1.5.

7.2.3.2 Diffusion Data

The large amount of data currentlyavailable on solute and self diffusion inCl-Zr was reviewed by Hood (1988). In gen-eral terms, single crystal, pure materialsshould be used for the measurements sothat there is some confidence that the re-sults represent intrinsic diffusion data. Fora review of diffusion in solids see Vol. 5,Chap. 2 of this Series.

The three main considerations in a studyof diffusion in Cl-Zr and Cl-Zr-based alloysare: the atomic size effect, the anisotropyof diffusion in the hexagonal lattice andthe effect of impurities.

Atom size: small solute atoms can dif-fuse through the interstitial sublattice atrates which may be 10 to 20 orders of mag-nitude faster than self or substitutional dif-fusion. Figure 7-14a (Hood, 1988) shows asummary of measurements of intrinsic dif-

fusion in Cl-Zr (generally extrapolatedfrom high temperatures), and made in sin-gle crystal specimens. The fast diffusingsmall solutes (H, Fe, Ni, up to 0 and N)are characterized by activation energiesbetween 0.6 and 2.5 eV and pre-exponen-tial factors of the order of 10- 7 to10- 4 m2 . S-1. Those elements are thoughtto be interstitial diffusers. Diffusion of thelarger solutes (Zr, Hf, Nb, Sn) has thecharacteristics of self- or substitutional dif-fusion, with activation energies of 2.8 to3.3 eV and preexponential factors of theorder of 10- 4 m 2 . S-1. The best estimatefor the activation energy of Zr self diffu-sion is 3.3 eV, of which 1.9 eV is the va-cancy formation energy and 1.4 eV is themigration energy (Hood et aI., 1992)When there is no experimental data, arough estimate of the migration energyand preexponential can be obtained fromthe size correlation curves shown in Fig.7-14 b.

Anisotropy: It has been recognized thatdiffusion along the a and c-axes in Zr isgenerally different. For that reason, inprinciple, diffusion measurements in Cl-Zrhave to be made in both directions. Inpractice this can be done more easily forthe fast solutes (Fe, Cr, Ni), for which ithas been determined that the diffusion co-efficients are maximum along the c-direc-tion, by a factor of D II /D -L = 3 (D II is thediffusion coefficient parallel to the c-axisand D -L is perpendicular to the c-axis). Forself diffusion the only measurement avail-able (Hood and Schultz, 1974) givesDII/D -L = 1.3. The diffusion of other slowdiffusers is al 0 nearly isotropic (Zhou,1992). The determination of such diffusionanisotropies is important: a factor of twoin the diffusion anisotropy has been show(Woo 1988) to produce significant effectsin sink biases and microstructural evolu-tion through the so-called DAD (diffusion


T(K)1000 800 700 600 500

/3 aL.......,

10-10

10-14

(U

10-180 (r

~- NI

III

.510-22

a

10-26

/ SUB

RE //

10-3Figure 7-14. (a) Self and solute diffusion

/data in a.-Zr (Hood, 1988). (b) Atom sizecorrelation for migration energy and pre-exponential factor for diffusion in a.-Zr(Hood, 1988).

1.0 1.2 1.4 1.6 1.8 2.0

(a) 103IT IK-1)

I I I

Ni 0 _- Zr(0 qo Fe -10 I- At .Ta

\ 3.0 ~ ~\ /~(u I

2.5 f- /12 I- \ oCr - .0 I

H~ \ /

'.. \ I~ \ - ~

I14 I- (u / • (r

0 \Fn ./

\ l AU '06 .£:. Ig " Ag II

AlI~~16 - 1.0 f- I -"'- oSb

I. FeNi I

O.S~

,.-

Zr(?) I18 '-

I I I I I I

0.12 0.14 0.16 0.12 0.14 0.16

(b) ELEMENT AL RADIUS (nml ELEMENT AL RADIUS (nml


anisotropy difference) effect, as mentionedin Sec. 7.3.1.4.

Impurities: The effect of impurities onZr self diffusion has only recently begun tobe understood. It was determined that sub-stitutional diffusion in Zr is dominated bythe effect of residual Fe. Hf diffusion innominally pure Zr ('" 50 ppm of Fe) wasshown to be higher by up to two orders ofmagnitude than Hf diffusion in ultrapureZr « 1 ppm Fe). In a conclusive experi-ment, when Fe was added to the ultrapurematerial, to the levels of the nominallypure material, the Hf diffusion coefficientsjumped up to the values observed in thenominally pure Zr (Hood et aI., 1992). TheFe-enhanced migration energy is 0.7 eV,in contrast to the intrinsic value of 1.4 eV.It is not known at present what kind ofdefect configuration and migration mecha-nism creates such a large diffusion en-hancement, e.g., a tightly bound vacancy-Fe complex having been proposed (Kinget aI., 1991).

7.3 In-Reactor Behavior

In this section, we examine the behaviorof Zr components when subjected to a nu-clear reactor environment.

In Sec. 7.3.1 the effects of the neutronflux in causing irradiation damage to theZr components are discussed. The effectsof this damage range from an increase inthe concentration of dislocation loops,causing hardening and loss of ductility andthe dimensional changes brought about byirradiation creep and growth, to secondphase dissolution, decomposition and re-precipitation, possibly influencing corro-sion resistance.

Waterside corrosion of Zr alloys is ex-amined in Sec. 7.3.2. In-reactor, high tem-peratures, irradiation radiolysis, and, pos-

sibly, deleterious changes in the materialcombine to accelerate corrosion rates thatare normally slow outside irradiation. Theeffects of alloying elements on corrosionrates are discussed and the special prob-lems of nodular corrosion, and hydrogen/deuterium pickup are considered.

In Sec. 7.3.3, the problem of stress cor-rosion cracking induced by the pellet-cladding interaction is discussed. This is atypical example of an in-reactor corrosionmechanism, since it requires claddingstresses induced by interaction with the ex-panding V0 2 pellet, while the corrodingagent is iodine originating from nuclearfissions.

7.3.1 Irradiation Damageand Irradiation Effects

7.3.1.1 Displacement Calculations

At the root of all irradiation effects dis-cussed later in this section is the displace-ment of atoms from their normal latticepositions by collisions with incident parti-cles. In the case of in-reactor behavior, weare concerned with the collisions of neu-trons with target atoms.

For the case of zirconium and 1 MeVneutrons, the recoil atom from the originalneutron-atom collision (called primaryknock-on atom, PKA), has a maximumenergy as high as 44 keV. This PKA de-posits its energy in a relatively localizedregion, causing secondary and higher or-der displacements. The localized nature ofthe energy deposition creates a region ofhigh displacement density, commonlycalled a collision cascade. The number ofdisplacements in a collision cascade causedby a PKA of energy E is v (E), typical val-ues being about 100 to 200.

The procedure for calculating v (E) andrelating it to the number of displacementsinduced in a material by a given neutron

fluence is well explained in the literature(Olander, 1976) and elsewhere in this Vol-ume (Chap. 9, Sec. 9.4.2), so it need notdetain us here. In order to perform thecalculation, the displacement energy Edneeds to be specified. This is the minimumenergy required to displace an atom fromits lattice site. According to measurementsin the high voltage electron microscope(HVEM) (Griffiths, 1987) Ed for the pro-duction of visible damage in pure Zr isabout 25 eV, varying by approximately1 eV with crystallographic orientation.This energy is much higher than theFrenkel pair formation energy, reflectingthe fact that the atom has to go through aDotential barrier in order to be displaced.

After being displaced from their moststable lattice position, the atoms have achoice of several interstitial configurationsto assume, of which the lowest energy onesare more likely to be found. These configu-rations have been studied by lattice calcu-lations using interatomic potentials thatpredict macroscopic properties of Zr withsome accuracy (Bacon, 1988). The possibledefect positions for the h.c.p. lattice areshown in Fig. 7-15. Calculations (Fuse,1985) show that the most stable interstitialis located at the Es position, and has anenergy of 3.83 eV.

Although the calculation for the conver-sion of neutron fluence to displacementsper atom has to be performed for eachspecific reactor, with its particular fluxcharacteristics and power history, a rea-sonable estimate of the conversion factor is2 displacements per atom (dpa) for each1025 n' m- 2 (E>l MeV). This means thatthe zirconium rods in a pressurized waterreactor (PWR) receive about 20 dpa in the

years they stay in the reactor, corre-sponding to about 10- 7 dpa . s -1.

Each atom is therefore displaced on theaverage about 20 times during the three

7.3 In-Reactor Behavior 25

(/)

x«I

()

1

Figure 7-15. Possible interstitial sites for h.c.p. zirco-nium. Letters in the figure indicate positions andtypes of interstitials (Fuse, 1985).

years it stays in reactor. Clearly, a lot ofannealing takes place concurrently, sincethe Zr cladding maintains its structural in-tegrity, and most of its good mechanicalproperties. Some of this annealing takesplace within the cascade itself, the numberof stable defects produced per displace-ment being much smaller than 1000/0 (Wooand Singh, 1992). However, the sheeramount of rearrangement that is requiredcreates a potential for microstructuralchanges under irradiation. The mecha-nisms by which this potential is translatedinto actual material changes under irradia-tion is reviewed in the following sections.

7.3.1.2 Irradiation Effects in the Zr Matrix

Under reactor operating temperatures,the point defects that escape immediate re-combination can migrate to sinks, such asgrain boundaries, free surfaces and dislo-cations. Due to the unequal bias of thedifferent sinks for the defects, vacanciesand interstitials can accumulate at differ-ent sinks, giving rise to macroscopic ef-fects. For example, when network disloca-tions are not present in large quantities(i.e., in recrystallized material), the pointdefects can agglomerate into dislocation


loops, causing a large increase in disloca-tion density.

Dislocation Structure

The as-fabricated material contains net-work dislocations, resulting from coldwork. Those are mixed in type, both

dislocations of the type 1/2 [0001] and1/2[1123] are found, which supports thegeneral idea of development of (c)-typedislocations under irradiation.

Voids

Contrary to the behavior of stainlesssteel, Zircaloy does not exhibit significantvoid formation under neutron irradiation(Farrell, 1980). Under electron irradiationswelling is observed in Zr which was previ-ously injected with helium (He) (Faulknerand Woo 1980). Accordingly, TEM stud-ies have not found many cavities and voidsin neutron or charged-particle irradiatedZr alloys. This was confirmed by Baiget al. (1989) who could not find any voidsin irradiated Zr by small angle neutronscattering. The reason is thought to be thatthe gases that could stabilize voids andcavities (O,H,N) are very soluble in the Zrmatrix, where their equilibrium concentra-tion can be quite high (Figs. 7-4 and 7-5).

Some cavities have nevertheless been~ound in crystal bar Zr after neutron irra-diation, in the interstitial denuded zonenear grain boundaries (Griffiths et al.,1988). This is probably due to the high


Figure 7-17. Di tribution ofvacancy and interstitial loopsclose to a grain boundaryin crystal-bar Zr irradiatedat 700 K to a fluence of1.1 x10 25 n'm- 2 (E>l MeV).Loops exhibiting inside con-trast with a 1212 diffractionvector are vacancy in nature

J (beam direction close to[1213]). Close to the grainboundary, vacancy loops formpreferentially, in contrast tothe bulk, where some inter-stitial loops also form (micro-graph courtesy of M. Griffiths,AECL, Chalk River).

(a)

(b)

Figure 7-18. (a) Formation of


vacancy concentration in those regions.For reasons unknown, voids are alsofound near second phase particles. Theirtotal number density is however quite lowand they are not thought to have mucheffect in the irradiation behavior of zir-conium alloys.

7.3.1.3 Irradiation Effectson Second Phases

Crystalline to Amorphous Transformation(Amorphization) of Precipitates

One of the most striking effects of irra-diation on Zr alloys is the crystalline toamorphous transformation (amorphiza-tion) observed in the intermetallic precipi-tates Zr(Cr, Fe)2 and Zr2(Ni, Fe), com-monly found in Zircaloys. Those pre-cipitates, described in Sec. 7.2.2, undergoamorphization under neutron irradiationas reported by Gilbert et al. (1985) andconfirmed by Yang et al. (1986). The trans-formation has been observed in Zircaloysirradiated at 550 to 620 K (from bothLWR fuel cladding and structural mate-rial) and at 350 K (calandria tubes fromCANDU reactors). At 350 K, both typesof precipitates are completely amorphousafter very low fluences (0.5 to 1 dpa). At thehigher temperature range, the Zr2(Ni, Fe)precipitates are completely crystalline,while the Zr(Cr, Fe)2 precipitates are par-tially amorphous having developed a "du-plex" structure, consisting of an amor-phous layer that starts at the precipitate-matrix interface, and gradually moves intothe precipitate until the precipitate is com-pletely amorphous. This is shown in Fig.7-19 A. A crystalline core is present asevidenced by the stacking fault contrast,while an amorphous layer has been formedthat will eventually envelop the whole pre-cipitate. Amorphization is associated witha depletion of iron from the amorphous

layer into the Zr matrix, while the Cr con-centration in the precipitate remains con-stant.

It is thought amorphization occurs be-cause the irradiated crystalline structure isdestabilized with respect to the amorphousphase due to the accumulation of irradia-tion damage. A review of the experimentalevidence and theoretical models for amor-phization of those precipitates under neu-tron and charged particle irradiation wasgiven by Motta et al. (1991) and Motta andLemaignan (1992).

Precipitate Dissolutionand Reprecipitation

After amorphization, precipitate disso-lution is accelerated, as illustrated in Figs.7-19 B, C, and D. A serrated interface isformed as the precipitates dissolve, indi-cating either a dissolution of the precipi-tate along preferential directions (Yang,1989), or a minimization of interfacial en-ergy by faceting.

After precipitate dissolution, and postirradiation annealing, Fe and Cr reprecipi-tate in the matrix, forming Cr-rich precipi-tates close to the original particle, and ironrich precipitates further away. This il-lustrates the difference in mobility ofthose two elements in Zr, mentioned inSec. 7.2.3.2. It is possible that some of theFe and Cr remains in solid solution.

Although precipitate dissolution may beaccelerated by the amorphous transforma-tion, it also happens under neutron irradi-ation at higher temperatures, where amor-phization does not occur, so amorph-ization is not a precondition for dissolu-tion.

There have been reports of irradia~tion induced precipitation and dissolutionof other types of precipitates in zirconiumalloys. ZrSnFe precipitates have been


Figure 7-19. Precipitateamorphization and dissolu-tion under neutron irradia-tion (courtesy of W. 1. S.Yang, GE). (A) The duplexstructure in a Zr(Cr, FE)2precipitate in Zircaloy-4 ir-radiated to 5 x 1025 n' m- 2

at 561 K, showing a crys-talline core surrounded byan amorphous layer. Ironis depleted in the amor-phous layer. (B), (C), and(D): Zr(Cr, Fe)2 precipitatedissolution along preferen-tial crystallographic direc-tions after irradiation to14.7x1025 n ·m- 2. The in-terface faceting is arrowedin (B).

found in Excel alloys (Zr-3.5 % Sn-1.00/0Nb-1.0 0/0 Mo) following neutron irradia-tion to 1.5 x 1026 n' m- 2 at 690 K (Grif-fiths,1988). This was paralleled by the ob-servation by Woo and Carpenter (1987) ofZrSSn 3 precipitates in Zircaloy-2 fol-lowing irradiation to 7.4 x 1024 n . m - 2(> 1 MeV) at 875 K (total dose about3 dpa). These last precipitates were laterfound to be associated with Fe. Also in2r-2.5 % Nb, the Fe-rich ~ phase loses itsFe to the a. phase, and extensive precipita-tion of fine-sized Nb rich precipitates inthe a. phase has been reported as a result(Coleman et aI., 1981). This is shown inFig. 7-20.

Outside irradiation, by contrast, the ~phase decomposes under thermal anneal-ing to a mixture of Zr-86 % Nb and theintermetallic (Zr Nb)3(Fe Cr).

Second phase redissolution and redistri-Jution of alloying elements, can have im-portant consequences for irradiationgrowth and corrosion resistance, as ex-plained below and in Sec. 7.3.2.

7.3.1.4 Irradiation Growth

Irradiation growth refers to the dimen-sional changes at constant volume of anunstressed material under irradiation (Fid-leris, 1988). For Zr single crystals, irradia-tion growth consists of an expansion alongthe a-direction, and corresponding con-traction along the c-axis.

In polycrystalline materials the situationis more complex, since grain boundariescan act as biased sinks for point defects, sothat grain shape and orientation play alarge role. Usually, however, growth be-havior of polycrystalline materials alsoconsists, of expansion along the a-direc-tion and contraction along the c-axis. Asnoted in Sec. 7.2.3.3, the fabrication pro-cess of Zr alloy components induces a tex-ture. For Zircaloy cladding tubes, prismplanes are preferentially aligned perpen-dicular to the axial (longitudinal) direc-tion, which means that irradiation growthcauses the axial length to increase and thecladding diameter and thickness to dimin-


Figure 7-20. ~-Nb precipi-tation in the a. phase ofExcel alloy (micrographcourtesy of R. W. Gilbert,AECL, Chalk River).

0.4 ,------------------,

Figure 7-21. Growth strain versus fluence for differ-ent amounts of cold work, at several irradiation tem-peratures from Rogerson (1988). The dashed linesshow the growth strain for cold-worked (CW) mate-rial: the strain is nearly linear with fluence and in-creases with temperature. The solid lines show thegrowth strain for annealed material: the growth ratesare noticeably smaller than in cold-worked material,and also increase with temperature. After the "break-away" fluence, growth rates in annealed materi .are comparable to those of cold-worked materia."Breakaway" growth in annealed materials has beenlinked to the development at that fluence of

fluence of about 3 x 1025 n . m - 2, there oc-curs the "breakaway phenomenon", thestrain rates jumping to approximately thevalues observed in cold worked materials(Fidleris, 1988). The observation of break-away growth has been linked to the devel-opment of


highly oriented grains. The parameters ad-justed to the behavior of Zircaloy werethen used to successfully predict thegrowth of Zr-2.5 % Nb (Holt and Fleck,1991).

in order to improve the knowledge of theparameters that control creep. Some ofthis work has been performed in materialstest reactors, and some during detailed ex-aminations of the behavior of structuralmaterial in power reactors. The experi-mental procedures used for those analyses,are based upon pressurized tube expan-sion stress relaxation measurements ofthin'plates, tensile testing or shear testing(i.e., springs loaded in tension). The lastmethod is a specific experiment designedto analyze the pure creep behavior. Forexample, Causey et al. (1984) were able toderive a contribution of about 30% (c +a)slip on the 1/3(11L3){10I1} system in ad-dition to the classical (a)-type basal slipby testing Zr-2.5 Nb in pure shear usinl.compressed helical springs or twisted tubesat 573 K. They also showed a weak depen-dence of creep rate on dislocation density.Irradiation creep in Zircaloy-2 is aniso-tropic and dependent on texture as shownby Harbottle (1978).

For practical purposes and for the lim-ited range of operating parameters, the ir-radiation creep strain rate may be de-scribed accurately with a simple equationof the form:

e=A . ( t)m. (Tn. e-Q/(RT) (7-2)

where the effect of flux, stress and temper-ature can be separated. The usual values ofthose exponents are m =0.6 to 1, n close to1. The fact that n is close to 1 means thatlarge creep strains can be sustained with-out failure. The activation energy, Q, islow, in the range of 5 to 15 kJ . mol-I, i.e.,0.05 to 0.16 eV' at- 1 (Franklin 1982).

The task of finding a mechanistic modelthat explains irradiation creep in Zr alloysis a daunting one, in view of the complexit.of the alloy its inherent single-crystal andtexture anisotropy and the effect of theirradiation field. For example, ince void

40001,.,,.,"uuu

//

/ t=2x1014 n.cm b.s-/v

/ v ~

/ ~ t_4X1013 n·cm-

// .. --- . -- _-- t=O-----If ..'.'.. 'o

o

0.2

7.3.1.5 Irradiation Creep

Irradiation creep refers to the slow de-formation under an external stress, experi-enced by a material under irradiation. Inanisotropic materials, such as Zr, creep hasalways to be separated from growth in ex-perimental situations. A convenient way todo this is to assume that creep and growthare linearly additive and define the irradia-tion creep strain as the additional strainthat results when the deformation processtakes place under an external stress. Thecreep strain is then the total strain minusthe growth strain.

As shown in Fig. 7-22, the creep defor-mation rate of Zr alloys is increased underneutron flux. Due to the objective of usingZr alloys in reactor, this phenomenon hasbeen subject to a large amount of experi-mental work, reviewed by Fidleris (1988),

0.6

2000 3000

time (hours)

Figure 7-22. Diametral creep of RX Zircaloy underinternal pressure (330 °e, CTH = 150 MPa). The irradia-tion affects both the primary and the secondary creep,and the creep rate is increased in proportion to thedose rate.

*-; 0.4c.~

U5

welling is practically nonexistent in Zr al-loys (Sec. 7.3.1.2), the standard rate theorymodel applicable to cubic metals couplingirradiation creep and void swelling (Brails-ford and Bullough, 1972), cannot be usedfor Zr alloys. In addition, many of the pa-rameters needed for mechanistic models,such as defect-defect interactions, are notknown for Zr, further complicating the en-deavor.

Matthews and Finnis (1988) recently re-viewed the literature on mechanisms of ir-radiation creep. It is thought that defor-mation during irradiation creep occurs bya combination of dislocation climb andglide, the climb being controlled by thestress-modified absorption of point defectsat dislocations. According to the so-calledSIPA (stress induced preferential absorp-tion) mechanism (Bullough and Willis,1975), dislocations that have their Burgersvectors parallel to the applied uniaxialstress, preferentially annihilate interstitialsthan vacancies, leading to dimensionalchanges due to the dislocation climb itselfand to the subsequent dislocation glide(Woo, 1979). Matthews and Finnis (1988),analyzing the possible origins for SIPAconclude that the elastodiffusion modelproposed by Woo (1984) is the strongestcandidate, since it is a first-order effect,involving the migration anisotropy of in-terstitials in an applied stress field.

The climb-assisted glide (I-creep) andSIPA mechanisms have recently been em-ployed by Woo et al. (1990) to derive ana-lytic expressions of their contribution tothe single crystal stress compliance tensorand used a self-consi tent model that treateach grain as an inclusion embedded in ananisotropic medium, for deriving expres-sions that relate the polycrystalline creepcompliance tensor with that of a singlecrystal. Using a self consistent model, anda numerical technique, and using the ana-


lytic expressions derived by Woo et al.(1990), Christodoulou et al.(1992) derivedthe relative contributions of basal, pyrami-dal and prismatic slip systems to the totalstrain in Zircaloy-2 and Zr-2.50/0 Nb. Itwas concluded that


o

1018 1019 1020 1021 1022

(a) Integrated neutron flux (E>1 MeV) (n cm-2)

7.3.1.7 Charged-Particle Irradiation

For the sake of completeness, the use ofcharged-particles (electrons and ions) forexperimental and theoretical studies of ir-

radiation effects in Zr alloys is discussed inthis section. The motivation for studyingirradiation effects with charged-particles isseveral fold: firstly, because their displace-ment rates are orders of magnitude higherthan under neutron irradiation, equivalentdoses in dpa are achieved correspondinglyfaster; secondly, the effect of experimentalvariables such as temperature, particletype and dose rate can be studied with rel-ative ease, and, finally, the irradiated sam-ples are usually not radioactive, makingthem much easier to handle.

In general, neither the results obtainedin such experiments have a one-to-one cor-respondence with the results from neutronirradiation nor should such close corre-spondence be expected in principle, as theirradiations are quite different. Electron ir-radiation differs from neutron irradiationboth in the absence of collision cascades,and in that the electron beam is focused,causing a much higher electron flux anddisplacement rate. For ion irradiation, theprincipal difference is that the rate of ionenergy loss per unit target thickness ismuch higher than that for neutron irradia-tion, which causes only a relatively thinlayer close to the surface to be irradiated,albeit at a much higher dose rate. Sinceirradiation effects are dependent on thebalance between irradiation damage andthermal annealing, any shift in the dis-placement rate can affect the observed ef-fects. Also, bulk effects of neutron irradia-tion, such as irradiation-induced creep andgrowth, and irradiation hardening are dif-ficult to study with near-surface irradia-tion. If, however, the characteristics ofeach type of irradiation are taken into ac-count, then charged-particle irradiationcan provide valuable information on the(mechanisms of irradiation damage and mi-crostructural evolution, which can be ap-plied to neutron irradiation.

Annealed

UTS

YS

Cold worked

UTS ::_:----------------------:::::::::::

YS

~30

c0

~ 200)c0a;~ 10-enc~

This increase in yield strength is associ-ated with a reduction in ductility, affectingboth the uniform elongation (through areduction of the strain hardening expo-nent), and the total elongation, decreasingfrom about 200/0 to 2-40/0 (Morize, 1984).This effect is clearly illustrated in Fig. 7-23from Price and Richinson (1978).

Most of the effects reviewed in Sec. 7.3.1have also been studied with charged-parti-cle irradiation, often with different results.Extensive studies of microchemical evolu-tion in Zircaloy-2 and Zircaloy-4 underproton irradiation have been performed byKai et ai. (1990, 1992). It was found thatproton irradiation may induce variationsin the chemical composition of intermetal-lic precipitates and increase the solute con-tent in the matrix. These changes appear toincrease the nodular corrosion resistanceof the alloy. Other investigations of theeffect of irradiation on uniform oxidationin Zircaloy-2 and 4 have been conductedby Pecheur et aI. (1992) and are reported inSec. 7.3.2.

Irradiation-induced segregation and pre-cipitation has been observed in at least twodifferent circumstances. Cann et aI. (1992)have observed that 3.6 MeV proton irradi-ation of annealed Zr-2.5 Nb at 720 K-770 K to 0.94 dpa causes a fine dispersionof Nb-rich platelike precipitates to appearwithin the r:x grains that is absent uponequivalent heat treatment of a controlsample, in agreement with the neutron ir-radiation results of Coleman et ai. (1981),reported in Sec. 7.3.1.3. Segregation of tinand precipitation as ~-Sn at the surface ofthin Zircaloy-2 foils has also been reportedafter 5.5 MeV proton irradiation to 1 dpaat 350 K (Motta et aI., 1992) a result thathas not been observed after neutron irradi-ation of calandria tubes at equivalent tem-peratures.

Amorphization of intermetallic precipi-tates in Zircaloy-2 and 4 under chargedparticle irradiation has been extensivelystudied, both experimentally (Motta et aI.,1989; Lefebvre and Lemaignan, 1989), and

eoretically (Motta and Olander, 1990).The critical temperature for amorphiza-tion of Zr(Cr,Fe)2 precipitates under elec-tron irradiation was found to be 300 K


lower than under neutron irradiation. ForAr ion irradiation, although the criticaltemperature was similar to that under neu-tron irradiation, the transformation mor-phology was quite different, amorphiza-tion no longer starting at the precipitatematrix interface.

As mentioned above, bulk effects suchas irradiation creep and growth and hard-ening, are more difficult to simulate withcharged particle irradiation, but some at-tempts have been made, especially inthe area of irradiation creep and growth.Parsons and Hoelke (1989) developed a"cantilever" beam method, which relatesthe irradiation-induced creep and growthof a cantilevered Zr beam to its deflectionwhen exposed on one side to 100 keV Neion irradiation. The results of the experi-ment differed from those of neutron irradi-ation, but no detailed modeling effort hasbeen undertaken as yet to rationalize thedifferences.

When both bulk and near-surface irradi-ation can be understood in terms of theatomic level mechanisms, then neutron ir-radiation can be simulated with chargedparticle irradiation. For example, forma-tion of voids under electron irradiationwas observed in a Zr sample that had beenpreviously charged with He (Faulkner andWoo, 1980) and dislocation loop forma-tion can be studied with electron irradia-tion (Woo et aI., 1992). This last study inci-dentally, supported the idea that micro-structural evolution in Zr alloys is affectedby the large diffusion anisotropy, as men-tioned in Sec. 7.2.3: irradiation-inducedvoids changed their shape under electronirradiation, preferentially shrinking in the


7.3.2 Corrosion Behavior

7.3.2.1 General Corrosion Behavior

Zr alloys are highly resistant to corro-sion in common media and are used forthat reason in the chemical industry (Tri-cot, 1989). Those alloys are however notimmune to oxidation and in the hightemperature water environment found ina power reactor (280 - 340°C at 10-15MPa), corrosion and hydriding controlsthe design life of fuel rods and other com-ponents. Several international meetingshave been organized to discuss this impor-tant design parameter, the latest ones hav-ing been reviewed by Franklin and Lang(1991).

In the early stages, a thin compact blackoxide film develops that is protective andinhibits further oxidation. This dense layerof zirconia is rich in the tetragonal al-lotropic form, a phase normally stable athigh pressure and temperature. As de-scribed in Fig. 7-24, the growth of the ox-ide layer thickness d, follows a power law,usually described by a quadratic relation-ship

d ex:. Jl (7-3)

The activation energy of 130 kJ . mol- i

(1.35 eV . at-i) for corrosion in the denseoxide regime (Billot et al., 1989) is equiva-lent to activation of the diffusion of oxy-gen in zirconia (Smith, 1969). Oxygen isconsidered to diffuse from the free surfaceof the oxide as 0 2 -, by a vacancy mecha-nism through the zirconia layer, and toreact with the zirconium at the matrix-ox-ide interface, but recent studies support thefact that those ions diffuse mostly throughgrain boundaries (Godlewski et a1., 1991).

For very thin oxide layers, the zirconiagrows in epitaxy on the metallic zirconi~m(Ploc, 1983). The Pilling- Bedworth ratIO,or ratio of the oxide volume to the parentmetal volume, is equal to 1.56 for zirconia.Thus as the oxide grows, the stress buildu.due to the volume expansion associatedwith oxidation induces a preferential oxideorientation that reduces the compressivestresses in the plane of the surface (Davidet al., 1971). This gives rise to various fi-brous textures. Those compressive stressesare one of the factors that explain the sta-bilization of the tetragonal form of zirco-nia in this layer. A chemical effect of thealloying elements could also be invoked toexplain that stabilization.

Figure 7-24. Uniform corrosionbehavior of Zirca10y-4 in water at633 K. The oxide thickness fol-lows a power law UP to a transi-tion (1.5 to 2 mm thickness) andthen remains linear. The metallur-gical state of the material affectsalso the corrosion rate (Claytonand Fisher, 1985).

500400200 300Time (days)

100

(Summary of 1600 data points)

200 ....-----.--.....---r-----r-,---r-""I""-01 ---r----,o Alpha annealed (RXA)180 0 stress relieved annealed (SRA) SR

~ 160 least square fitted line'"0 2or § data spread around averagea, 140E-120c'g,100

-§, 80'Qj

~ :~!~~Oo

As the oxidation proceeds, the compres-sive stresses in the oxide layer cannot becounterbalanced by the tensile stresses inthe metallic substrate and plastic yield inthe metal limits the compression in the ox-ide. The tetragonal phase becomes un-stable and the oxide transforms to a mon-oclinic form (Godlewski et aI., 1991). Thismartensitic-type transformation is associ-ated with the development of a very fineinterconnected porosity that allows the ox-idizing water to access closer to the corro-sion interface (Cox, 1969). The size ofthose pores has been measured to be verysmall: Cox (1968) has used a modified mer-cury porosimeter to determine pore sizessmaller than 2 nm. His work has been con-irmed by nitrogen absorption kinetics to

pore sizes smaller than 1 nm, correspond-ing to a volume fraction of the order of1 0/0 (Ramasubramanian, 1991). Once thistransition has occurred, only a portion ofthe oxide layer remains protective. Thecorrosion kinetics are therefore controlledby diffusion of oxygen only through thedense protective oxide layer next to themetal substrate. Since the thickness of thislayer remains constant, in the range of1 Jlm (Garzarolli et aI. 1991) the corro-sion rate is constant after this transition(Fig. 7-24).

In this dense oxide layer the structure ofthe zirconia, which controls the posttransi-tion corrosion kinetics, is complex and stillunder discussion. Starting from the metal-oxide interface, a very thin layer of amor-phous oxide has been reported under par-ticular conditions, about 10 nm in thick-ness (Warr et aI., 1991). The existence ofepitaxy shows that this layer, if present,should not be continuous. It is followed by

zone of very small zirconia crystallites,10-20 nm, that become larger in diameterand columnar in shape further into the ox-ide layer (Bradley and Perkins, 1989).


For thick oxide layers, in excess of50 Jlm, the oxide may spall, leaving zirco-nia particles free to flow in the coolingwater, and giving rise to a much thinneroxide film. On the one hand, this processcould be beneficial since it reduces themetal-oxide interface temperature, whichis the parameter controlling oxidation ki-netics. However, design considerations onremaining cladding thickness after oxida-tion does not allow for massive spalling incommercial reactors. Other undesirableconsequences of spalling are the buildup ofactivity in the coolant and safety consider-ations, since those particles can interferewith the functioning of valves.

In BWRs, as shown in Fig. 7-25, nodu-lar corrosion is the limiting design consid-eration. This behavior, specific to the boil-ing water reactor, can be reproduced bytesting alloys in steam at 500°C (Schemel,1987). Several mechanisms have been pro-posed for the nucleation of those nodules,leading to various possible sites for nodulenucleation: metallic matrix grain bound-aries, local rupture of the continuous denseoxide at an early stage of growth, localvariations in composition and precipitatedensities or crystallographic orientation ofclusters of grains (Charquet et aI., 1989 b;Ramasubramanian, 1989). As the corro-sion progresses the nodules grow in sizeand thickness and their number densityincreases leading to a complete coverageof the metal. The ~ quench structure de-scribed in Sec. 7.2.2.2 significantly im-proves the resistance of Zircaloy-4 tonodular corrosion, however as higher bur-nups are reached uniform corrosion couldthen become a problem.

7.3.2.2 Oxidation of the Precipitates

Since most of the metallic alloying ele-ments pre ent in Zircaloys are added for


(a)

(b)

(c)

Figure 7-25. (a) Typical aspect of nodular corrosionof Zircaloy-4 obtained in a 500°C steam test. (b)Higher magnification of a cracked nodule in Zirca-loy-4, obtained during a 10.3 MPa steam test at500°C for 25 h, showing the extensive oxide crackingassociated with the process. In addition to circum-ferential cracks in the flanks of the nodules, a verticalcrack can also be seen running between the nodules[Courtesy of NFIR (Nuclear Fuel Industry ResearchGroup)]. (c) Greater detail of the nodule, showing thatthe upper portion of the nodule consists of a series ofparallel, cracked, oxide layers (Courtesy of NFIR).

improving corrosion resistance, the mech-anism of their interaction with the oxida-tion front is of great importance. Due tothe fine structure of the precipitates nospecific observation of this process wasavailable until recently, when a few studieshave been performed using advancedSTEM. It was shown that the precipitatesare incorporated in metallic form into theoxide layer, and oxidize only afterwards,deeper into the oxide layer. In particular,iron has been shown to remain unoxidizedin the dense oxide layer (Garzarolli et al.,1991; Pecheur et al., 1992). The precipi-tates slowly release part of their iron in theoxidized matrix so oxide chemistrychanges as the oxidation proceeds. It isfound that the oxidation of iron coincidewith the oxide transition from tetragonalto monoclinic.

Also, some of the crystalline intermetal-lic precipitates are found to be amorphousafter incorporation into the oxide layer(Pecheur et al., 1992). This transformationis not caused by irradiation since those arealso observed in nonirradiated oxidizedmaterial. The crystal-to-amorphous trans-formation in the oxide layer could becaused by changes in chemistry, notablyhydrogen intake (the nearest neighbor dis-tance in the amorphous phase is biggerthan in the corresponding precipitatesamorphized by irradiation in the Zr ma-trix).

The effect of the release of iron in theoxide is still under consideration. A corre-lation has been found between the oxida-tion of the precipitates and the tetragonal-to-monoclinic transformation of the zirco-nia, but without clear knowledge of itsorigin (Godlewski et al., 1991). A chemicaleffect on the stabilization of the tetragonphase of the oxide may be present.

An additional point is the role played bythe precipitates as possible short circuits


*

0.8

*

*

0.1

o

o

,I

*,*1** II

// 500°C/16h

/

~/ *o it ..., ......g> ----'

o *0

out-at-pile

3

2 2oL..

10

co'inoL..o 0.02u

~ 30


gen to the cooling water. Then a generalreduction of stable radiolytic species oc-curs and a reduction in corrosion rate ofthin oxide ensues. However this effect isnot as efficient in boiling conditions com-pared to pressurized ones, due to the segre-gation of Hz in the steam phase (Ishigureet aI., 1987).

Radiolytic enhancement of corrosioncan also occur in the case of PWR's, inpost transition thick oxide films, or duringcorrosion testing of coupon specimens in areactor corrosion loop. There, although nosteam is expected to be present, the waterchemistry in the pores of the thick oxideshould be much like that present in BWRs,i.e., a two-phase regime (Johnson, 1989).

Some particular cases of localized corro-sion enhancement have recently beenlinked to an increase in metastable oxidiz-ing species due to ~- radiolysis. The irradi-ation enhancement of corrosion is well ob-served for thick oxide films, where a three-to four-fold enhancement is commonlyseen (Marlowe et aI., 1985).

The reported occurrences are character-ized by the presence of dissimilar materialsin the vicinity of the corroding material.As recently reviewed, enhanced corrosionhas been observed in front of stainlesssteel, copper cruds, platinum inserts and inthe case of gadolinia bearing poison rods(Lemaignan, 1992). In all the reportedcases, strong ~ emitters are present and thelocal energy deposition rates of those par-ticles within the coolant are much largerthan the bulk radiolysis contribution dueto the neutrons and gammas. Thus, inreactor, the local intensity of the radiolysishas to be considered, as it will change thelocal chemistry for alloy corrosion.

In addition to the radiolysis describedabove, corresponding to chemical evolu-tion by reaction in the bulk, specific con-siderations should be given to the fact that

the pores have a large surface to volumeratio, leading to additional reactions at thesurface of the pores, instead of a recombi-nation between them as in the bulk. Forinstance, the HzOz molecules can be ad-sorbed on the surface of the pores leadingto the reaction:

HzOz ~ HOads . +HOads .

instead of recombining in two steps with2 H as the standard back reaction to waterafter radiolysis. This leads to a higher oxy-gen potential at the surface of the ZrOz ,giving a higher corrosion rate.

7.3.2.4 Hydrogen Pickup

During oxidation of Zr alloy compo-nents in reactors or in autoclaves, the re-duction of water by the Zr alloy followsthe general reaction scheme:

2HzO+Zr ~ ZrOz +4H'

As described above, the reaction pro-ceeds in several steps and the oxidationprogresses at the metal-oxide interface bydiffusion of Oz - ions through the oxide.The reduction of the water molecules atthe coolant-oxide interface releases hy-drogen as radicals H+. They are chemi-cally adsorbed at the tips of the oxidepores and their evolution controls thebehavior of the hydrogen. Most of themrecombine, creating hydrogen moleculesthat escape through the pore and dissolveinto the coolant. A limited amount caningress in the oxide and diffuse through tothe metallic matrix and then react with Zrfor the formation of hydrides, when theterminal solid solubility is exceeded. Cor-rosion experiments performed using tritium-doped water have shown that the Hdissolved in the coolant is not trapped bythe matrix, but that radicals obtained dur-

ing the reduction of water are necessary forhydrogen pick-up (Cox and Roy, 1965).

The diffusion coefficient of H in Zr0 2has been measured with difficulty becauseof its very low value and because of thecontribution of grain boundary diffusionand surface diffusion to diffusion in thepores of the oxide. Recent measurementsby Khatamian and Manchester (1989)have confirmed earlier results in the rangeof 10- 17 to 10- 15 m 2 . S-1 at 400°C, de-pending on the condition of measurementsand chemical composition of the alloy onwhich the oxide is grown. With such lowvalues, the penetration of H in the oxide islimited, and the zirconia layer is indeedorotective with respect to H ingress.

Care should be taken not to introducecatalyzers of the hydrogen molecule disso-ciation reaction that could enhance Hpick-up (or hydrogen uptake) - i.e., thefraction of the hydrogen produced by re-duction of water that is trapped into the Zralloy.

new materials science andtechnology · 2015. 9. 1. · reprint from materials science andtechnology...

Documents